The disclosures of all patents and publications cited in the specification are incorporated herein by reference.
DNA polymerase III holoenzyme ("Pol III") was first purified and determined to be the principal replicase of the E. coli chromosome by Kornberg (A. Kornberg, 1982 Supplement to DNA Replication, Freeman Publications, San Francisco, pp 122-125). The E. coli replicase is composed of a DNA polymerase subunit accompanied by multiple accessory proteins and contains at least ten subunits in all (McHenry and Kornberg, 1977, J. Biol. Chem., vol. 252, pp 6478-6484; Maki and Kornberg, 1988, J. Biol. Chem., vol. 263, pp 6551-6559). It has been proposed that chromosomal replicases may contain a dimeric polymerase in order to replicate both the leading and lagging DNA strands concurrently (Sinha et al., 1980, J. Biol. Chem., vol. 225, pp 4290-4303).
One of the features of Pol III which distinguishes it as a chromosomal replicase is its use of ATP to form a tight, gel filterable "initiation complex" on primed DNA (Burgers and Kornberg, 1982, J. Biol. Chem., vol. 257, pp 11468-11473). The holoenzyme initiation complex completely replicates a uniquely primed bacteriophage single-strand DNA ("ssDNA") genome coated with the ssDNA binding protein ("SSB"), at a speed of at least 500 nucleotides per second at 30.degree. C without dissociating from an 8.6 kb circular DNA even once (Fay et al., 1981, J. Biol. Chem., vol. 256, pp 975-983; O'Donnell and Kornberg, 1985, J. Biol. Chem., vol. 260, pp 12884-12889; Mok and Marians, 1987, J. Biol. Chem., vol. 262, pp 16644-16654). This remarkable processivity, i.e., the high number of nucleotides polymerized in one template binding event, and catalytic speed is in keeping with the rate of replication fork movement in E. coli, 1 kb/second at 37.degree. C. (Chandler et al., 1975, J. Mol. Biol., vol. 94, pp 127-131).
Within Pol III, the .alpha. subunit (dnaE) contains the DNA polymerase activity (Blanar et al., 1984, Proc. Natl Acad. Sci. USA, vol. 81, pp 4622-4626), and the .epsilon. subunit (dnaQ,mutD) is the proofreading 3'-5' exonuclease (Scheuermann and Echols, 1985, Proc. Natl Acad. Sci. USA, vol. 81, pp 7747-7751; DeFrancesco et al., 1984, J. Biol. Chem., vol. 259, pp 5567-5573). The .alpha. subunit forms a tight 1:1 complex with .epsilon. (Maki and Kornberg, 1985, J. Biol. Chem., vol. 260, pp 12987-12992). Whereas most DNA polymerases have 3'-5' exonuclease activity, only the holoenzyme relegates this activity to an accessory protein. The following three accessory proteins of the holoenzyme are known to be required for DNA replication as they are products of genes that are essential for cell viability: .beta. (dnaN) (Burgers et al., 1981, Proc. Natl Acad. Sci. USA., vol. 78, pp 5391-5395), .tau. and .gamma. (the latter two both encoded by the dnaXZ gene) (Kodaira et al., 1983, Mol. Gen. Genet., vol. 192, pp 80-86).
Important to the assessment of the individual functions of accessory proteins has been the availability of subassemblies of Pol III. Subassemblies include Pol III *, which is the holoenzyme lacking only .beta. (McHenry and Kornberg, 1977, J. Biol. Chem., vol. 252, pp 6478-6484); Pol III core, a heterotrimer of .alpha..epsilon..theta. that contains the DNA polymerase .alpha. subunit and the proofreading 3'-5' exonuclease .epsilon. subunit (McHenry and Crow, 1979, J. Biol. Chem., vol. 254, pp 1748-1753; Maki and Kornberg, 1985, J. Biol. Chem., vol. 260, pp 12987-12992; Scheuermann and Echols, 1985, Proc. Natl Acad. Sci., vol. 81, pp 7747-7751); Pol III', a dimer of .alpha..epsilon..theta..tau. subunits (McHenry, 1982, J. Biol. Chem., vol. 257, pp 2657-2663); the .gamma. complex, .gamma..sub.2 .delta..delta.'.chi..psi., composed of 5 accessory proteins (Maki, and Kornberg, 1988, J. Biol. Chem., vol. 263, pp. 6555-6560); and .gamma..chi..psi. complex (O'Donnell, 1987, J. Biol. Chem., vol. 262, pp 16558-16565). Due to the low abundance of the holoenzyme in cells, these subassemblies have hitherto been available only in microgram quantities.
The core polymerase has weak catalytic efficiency and is only processive for approximately 11 nucleotides (Fay et al., 1981, J. Biol. Chem., vol. 256, pp 976-983). The catalytically efficient holoenzyme can be restored upon mixing core with both the .beta. and the .gamma. complex (Wickner, 1976, Proc. Natl Acad. Sci. USA, vol. 73, pp 3511-3515). Reconstitution of the holoenzyme proceeds in two stages. In the first stage, the .gamma. complex and the .beta. subunit hydrolyze ATP to form a tightly bound "preinitiation complex" clamped onto the primed DNA. In the second stage, the preinitiation complex binds the core and confers onto it highly processive synthesis. ATP is only required in the first stage.
The .gamma. complex both recognizes primed DNA and hydrolyzes ATP to clamp the .beta. subunit onto DNA. In fact, one .gamma. complex molecule can act catalytically to form many .beta. clamps on multiple DNA molecules (Stukenberg et al., 1991, J. Biol. Chem., vol. 266, pp 11328-11334). The .gamma. complex therefore has the characteristics of a chaperonin. Namely, it acts catalytically to couple ATP to assembly of a complex (".beta..cndot.DNA"). Only the .gamma. and .delta. subunits are required to clamp .beta. onto primed DNA (O'Donnell, 1987, J. Biol. Chem., vol. 262, pp 16558-16665).
The holoenzyme Pol III is purified from E. coli as a multiprotein particle (O'Donnell, 1992, Bioessays, vol. 14, pp 105-111). The probable orientations of the subunits within the holoenzyme can be deduced from the known interactions among subunits.
Both .gamma. and .tau. are produced from the same dnaXZ gene. The .gamma. subunit is produced by a frameshift event which occurs after approximately two thirds of the gene has been translated (Tsuchihashi and Kornberg, 1990, Proc. Natl Acad. Sci. USA, vol. 87, pp 2516-2520; Flower and McHenry, ibid., pp 3713-3717; Blinkowa and Walker, 1990, Nuc. Acids Res., vol. 18, pp 1725-1729). The frameshift is followed within two amino acids by a stop codon. The .tau. subunit is the full length product of the dnaXZ gene. Approximately equal amounts of .tau. and .gamma. are produced in E. coli (Kodaira et al., 1993, Mol. Gen. Genet., vol. 192, pp. 80-86).
One of the roles of the .tau. subunit is to serve as a scaffold to dimerize the polymerase subunits. One indication that .tau. dimerizes the polymerase is from the purification and characterization of the 4-protein (.alpha..epsilon..theta..tau.) subassembly called Pol III'. Pol III' appears to be a dimer of all four subunits (McHenry, 1982, J. Biol. Chem., vol. 257, pp 2657-2663). Since the .alpha..epsilon..theta. Pol III core appears to contain only one of each subunit, the dimeric structure of Pol III' is believed to be due to the .tau. subunit. Study of pure .alpha. and .tau. subunits has shown the isolated .alpha. subunit, i.e., the polymerase is only a monomer, even at high concentration. However, the .tau. subunit, which is a dimer (Tsuchihashi and Kornberg, 1989, J. Biol. Chem., vol. 264, pp 17790-17795), binds two molecules of .alpha.. Hence, .tau. appears to be the agent of polymerase dimerization. The .tau. subunit also increases the affinity of the core polymerase for the preinitiation complex (Maki and Kornberg, 1988, J. Biol. Chem., vol. 263, pp. 6561-6569) and is a DNA-dependent ATPase, although the function of its ATPase activity is unknown (Lee and Walker, 1987, Proc. Natl Acad. Sci. USA, vol. 84, pp 2713-2717). The .alpha..epsilon..theta. core polymerase appears to form a dimer when it is sufficiently concentrated. Since a 1:1 complex of .alpha..epsilon. shows no tendency to dimerize to (.alpha..epsilon.).sub.2, the .theta. subunit has also been proposed to aid polymerase dimerization (Maki et al., 1988, J. Biol. Chem., vol. 263, pp 6570-6578). The .beta. subunit also interacts with the core, specifically by direct interaction with .alpha. (O'Donnell, 1987, J. Biol. Chem., vol. 262, pp 16558-16565; LaDuca et al., 1986, J. Biol. Chem., vol. 261, pp 7550-7557). The .gamma. subunit does appear to bind the polymerase subunit, implying it is the C-terminal portion of .tau. (lacking in .gamma.) that binds .alpha..
In determining the arrangement of the 5 subunits of the complex, isolation of a .gamma..chi..psi. complex suggests .chi., .psi., or both bind directly to .gamma.. Neither .chi. nor .psi. have been found in association with .tau.. Possibly in .gamma., the omission of the C-terminal portion of .tau. yields a surface unique to .gamma. which specifically binds .chi. and .psi.. Although it is not clear how the .gamma. complex contacts the polymerase, the .tau. subunit can interact in vitro with either .delta. or .delta.' to form a complex that can lock .beta. onto primed DNA (O'Donnell, 1987, J. Biol. Chem., vol. 262, pp 16558-16565). This suggests that .tau. contacts the .gamma. complex by interaction with .epsilon. and .epsilon.'.
DNA polymerases that duplicate chromosomes are remarkably processive multiprotein machines. These replicative polymerases remain in continuous association with the DNA over tens to hundreds of kilobases. The high processivity of Pol III holoenzyme lies in a ring-shaped protein that acts as a sliding clamp for the rest of the machinery. A ring-shaped sliding clamp is likely to be general for replicative polymerases, being formed by the PCNA protein of yeast and humans and the gene 45 protein of the T4 phage.
The subunit structure of the E. coli replicative polymerase is presented in Kornberg and Baker (Kornberg and Baker, 1991, DNA Replication, W. H. Freeman, New York, pp 165-207). During purification, this multiprotein polymerase generally dissociates into three components: a DNA polymerase, a multisubunit accessory protein complex, and a single subunit accessory protein. The DNA polymerase components of E. coli are Pol III core, which is a heterotrimer of .alpha. (polymerase), .epsilon. (3'-5' exonuclease), and .theta.. Prokaryotic polymerases are not highly processive on their own; for this they require their respective accessory proteins. The accessory protein complex of E. coli is the five subunit .gamma. complex, a DNA dependent ATPase that recognizes a primer-template junction. The single accessory subunit of E. coli is the .beta. subunit, which has no inherent catalytic activity, but stimulates the ATPases activity of the accessory protein complex.
The E. coli accessory proteins, .gamma. complex plus .beta. subunit, hydrolyze ATP to form a tight protein clamp (termed a "preinitiation complex") on primed single-stranded (ss) DNA coated with ssDNA binding protein (Wickner, 1976, Proc. Natl Acad. Sci. USA, vol. 73, pp 3511-3515; O'Donnell, 1987, J. Biol. Chem., vol. 262, pp 16558-16565; Maki and Kornberg, 1988, J. Biol. Chem., vol. 263, pp 6561-6569). The core polymerase then associates with the preinitiation complex to form the highly processive holoenzyme. The preinitiation complex thus acts as a gliding clamp on primed DNA.
In the E. coli system, the .gamma. complex can be separated from the preinitiation complex by gel filtration, leaving only a dimer of .beta. on primed DNA. This dimer retains the capacity to confer highly processive synthesis onto the core polymerase (Stukenberg et al., 1991, J. Biol. Chem., vol. 266, pp 11328-11334). Hence the .gamma. complex is not required during processive elongation. The role of the .gamma. complex is to recognize primed DNA and to couple ATP hydrolysis to clamp .beta. onto DNA (Onrust et al., 1991, J. Biol. Chem., vol. 266, pp 21681-21686). Only the .gamma. and .delta. subunits are essential to clamp .beta. onto DNA, although the other subunits of the .gamma. complex modulate the activity of the .gamma. and .delta. subunits (O'Donnell and Studwell, 1990, J. Biol. Chem., vol. 265, pp 1179-1187).
Several .beta. clamps can be transferred by one .gamma. complex onto a single circular plasmid DNA molecule containing one nick (Stukenberg et al., op. cit.). These .beta. clamps freely diffuse along the plasmid DNA and will slide off the end of this DNA if the circular plasmid is cut with a restriction enzyme. The sliding motion of .beta. is bidirectional, independent of ATP, and occurs only on duplex DNA (Stukenberg et al., op. cit.). It is known that .beta., once transformed by the .gamma. complex into a strong DNA binding protein, readily slides off linear DNA. Hence it was hypothesized that the .beta. dimer is shaped like a ring that encircles the duplex like a doughnut (Stukenberg et al., op. cit.). This .beta. ring can then slide off linear DNA like a washer off the end of a steel rod.
The .beta. clamp confers processivity onto the core polymerase by binding directly to the polymerase to subunit, thereby tethering the polymerase to DNA for processive syntheses. As the polymerase extends the 3' terminus, it simply pulls the .beta. sliding clamp along.
X-ray analysis of .beta. shows that it indeed has the shape of a ring (Kong et al., 1992, Cell, vol. 67, pp 425-437). The central cavity is of sufficient diameter to accommodate duplex DNA and is lined with 12 .alpha. helices that are supported by a single continuous layer of .beta. sheet structure all around the outside. Although the .beta. ring is uninterrupted and looks like a single molecule, it is actually a head-to-tail dimer. The width of the .beta. clamp is approximately that of one turn of the DNA helix, and the head-to-tail arrangement creates physically distinct front and back faces. The "A" face is rather flat and is more negatively charged than the "B" face, which is characterized by several loops. The .gamma. complex likely opens and closes the .beta. ring around DNA and orients the correct face toward the primer terminus for interaction with the polymerase. Each .beta. monomer is composed of three domains that have the same three-dimensional structure, giving the .beta. dimer a six-fold repeat appearance. This high degree of symmetry in the .beta. ring could help promote smooth gliding along the symmetrical DNA duplex.
The .beta. clamp is negatively charged overall, but the .alpha. helices lining the central cavity are positively charged, perhaps to stabilize the .beta. clamp on DNA. Ionic interaction between .beta. and DNA must be mediated by water because the inside diameter of the .beta. ring is too large for direct contact between the basic side chains and the DNA at the center. The lack of direct contact between .beta. and DNA may facilitate the sliding motion.
It is not clear why the clamp is separate from the polymerase, but it seems likely that there are advantages in having a separate clamp. For example, the lagging strand is synthesized as a series of short fragments, and in E. coli the polymerase must complete a fragment then dissociate from it and reassociate with a new primer made by primase all within a second. Biochemical studies show that the polymerase core rapidly dissociates from its .beta. clamp, but only after DNA synthesis is complete, and then it reassociates with another .beta. clamp on a new primed template (O'Donnell, 1987, J. Biol. Chem., vol. 262, pp 16588-16565; Studwell et al., 1990, UCLA Symp. Mol. Cell Biol. New Ser., vol. 127, pp 153-164). Hence, rapid disassembly/reassembly of the polymerase with separate .beta. clamps may provide a mechanism by which a highly processive polymerase can rapidly recycle on and off DNA during syntheses of the numerous lagging strand fragments.